[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [microblaze-uclinux] FS-Boot, EDK 10.1 Spartan 500ERevD



ut Michal
    My problem seems to be a timing issue.  I am not sure how the differences between EDK 9.1 and 10.1 change the design but I can boot successfully with new FS-boot out of Bram if I load it from the SDK after I previously load the FS-boot from the PreBuilt.  The new U-Boot and Kernel were flashed using the pre-built FS-boot  but using the config and images from the new build.  The "Hardware design" is loaded by the respective FS-boots.  The software in FS-boot has not been changed.  Another thing I have noticed is that the 6.0 Microblaze uses 10 bram blocks to implement two 8K cache blocks where the &.1 Microblaze uses the 10 bram blocks to implement two 4K cache blocks.  Something strange there.  I think the problem is caused by the Xilinx IP's.
Thanks
Hubert

Michal Simek wrote:
Hi Hubert,

FS-boot is in bram - then looks at specific place in flash where is stored
kernel image or u-boot image (only relocable format is acceptable) -> if success
just jump on specific address. Memory map and specific address is counted by
BSP.(Size of memory could be doubled but this shouldn't be a problem -> uncached
shadow case). If you have doubled size memory size, please set correct value in
 main_memory_size in BSP(EDK-> software setting)
In your case seems your flash starts at 0x21000000. Firsts sectors are used by
U-BOOT and then is space for linux kernel - (image.ub - created by mkimage
tool). Because you U-BOOT is set "only" by ub.config.img which set your U-BOOT
variable there is one default choice which is called bootcmd. Please write pri
in U-BOOT. you should see it. There could be written something like bootm
0x210c0000 and variable bootdelay should be set to 5. This means that after
U-BOOT startup you will wait 5 seconds and then if you don't interrupt it, you
will run bootcmd command. BOOTM command just look to your specific address
(0x210c0000 in your case) and look at magic sequence which have every mkimage
formats. If is found, just follow U-BOOT code and copy and jump code to ram. If
not U-BOOT announce "Bad Header Checksum" error because U-BOOT seems to counted
crc of image and of course is bad. Correct crc value is in image header.
This concept is prepared for platform where u-boot run prepared image in flash.
If you want to use different values you can just change them.

For testing is good to set booting from tftp server
setenv bootcmd tftp 0x27080000 image.ub\;bootm
saveenv

For disabling 5 sec delay
setenv bootdelay 0
saveenv

U-BOOT is more powerful than fs-boot especially for testing. On u-boot mailing
list you can find some discuss about how is possible to do more sophisticated
booting strategy or how to safely upgrade kernel image on your running platform.
On the other hand fs-boot is in bram and if is smaller and faster for kernel
running. This shorten kernel reboot. There is a lot of issues which every
booting strategy brings.

I hope you have better look how this work and why.

Regards,
Michal Simek


Hubert Bahr wrote:
  
After downloading the image the new FS-boot is going to the 0x27fc0000
address (as reported by the debugger with the breakpoint set just before
the GO routine), which is the same address reported from the U-boot
booted by the prebuilt FS-boot.   All The addresses reported during
startup from linux reported when starting with the prebuilt FS-boot
agree with those listed in the XPS address map and listed in the
auto-config.h file so I am assuming that address mismatch is not the
problem.  This now leaves me with some hardware mismatch that wasn't
picked up during the bit mask generation.
When I run from the breakpoint It then reports
SDRAM :
       Enabling caches :
               Icache:OK
               Dcache:OK
       U-Boot Start:0x27fc0000
       Malloc Start:0x27f60000
       Board Info Start:0x27f5ffd0
       Boot Parameters Start:0x27f4ffd0
FLASH: 16 MB
ETHERNET: MAC:00:0a:35:00:22:01

Hit any key to stop autoboot:  0
## Booting image at 210c0000 ...
Bad Header Checksum
U-Boot>
This is strange since that is in flash and when i boot from the prebuilt
FS-boot it passes this point and continues to boot.  It is starting to
look like it is having problems reading the Flash chip.   Any Ideas?

Hubert Bahr wrote:
    
I guess what you are implying is that Since FS_boot is built inside
the XPS and U_Boot etc are built outside they don't refer to the same
addresses.  I thought that the autoconfig script would take care of
that?  Where would I check that?
Hubert

Olof Backing wrote:
      
On 11 okt 2008, at 08.26, Hubert Bahr wrote:

        
 Starting from the Petalogix reference design for the Spartan 500E
Rev D Starter Kit I can't get to Uboot from the FS-Boot.   The
pre-built images work fine.  The new images I built work fine with
FS-boot image from the pre-built.  But it will not boot either
U-boot in the flash or a new one downloaded via Kermit.  Either way
it hangs after the
FS-BOOT:   Booting image...
response.

As far as I can tell the only changes from the reference are the
ones that XPS generated in the upgrade from 9.1.  Any suggestions as
to my next step will be appreciated
          
Have you checked the baseaddress of the uartlite (or similar)?

/Olof
        
___________________________
microblaze-uclinux mailing list
microblaze-uclinux@xxxxxxxxxxxxxx
Project Home Page : http://www.itee.uq.edu.au/~jwilliams/mblaze-uclinux
Mailing List Archive :
http://www.itee.uq.edu.au/~listarch/microblaze-uclinux/

      
___________________________
microblaze-uclinux mailing list
microblaze-uclinux@xxxxxxxxxxxxxx
Project Home Page : http://www.itee.uq.edu.au/~jwilliams/mblaze-uclinux
Mailing List Archive :
http://www.itee.uq.edu.au/~listarch/microblaze-uclinux/


    
___________________________
microblaze-uclinux mailing list
microblaze-uclinux@xxxxxxxxxxxxxx
Project Home Page : http://www.itee.uq.edu.au/~jwilliams/mblaze-uclinux
Mailing List Archive : http://www.itee.uq.edu.au/~listarch/microblaze-uclinux/

  

___________________________ microblaze-uclinux mailing list microblaze-uclinux@xxxxxxxxxxxxxx Project Home Page : http://www.itee.uq.edu.au/~jwilliams/mblaze-uclinux Mailing List Archive : http://www.itee.uq.edu.au/~listarch/microblaze-uclinux/